The rostral migration of epidurally administered morphine through the cerebrospinal fluid (CSF) into the brain is believed to be related to the hydrophilic properties of morphine and responsible for some of its adverse effects such as "late respiratory depression" . After lumbar epidural administration Gourlay et al. detected proportionally less meperidine compared to morphine in cervical CSF  and after fentanyl administration peak cervical CSF concentrations were only 10% of those in the lumbar region . By a rapid diffusion of lipophilic opioids into the lipid-rich areas of the spinal cord the amount that is available for cephalad migration via passive CSF flow would be effectively reduced. Consistent with these pharmacokinetic properties epidurally administered lipophilic opioids, such as sufentanil, produce rapid and effective pain relief , but less likely "late respiratory depression" would occur. Compared to plasma, the pharmacokinetics in CSF more closely should separate epidural from intravenous (IV) opioids and this property necessitates investigation in both CSF and plasma .
The purpose of this study was to determine pharmacokinetic variables in both lumbar CSF and plasma after an epidural bolus dose of sufentanil, and to examine first, the rate of dura penetration and the availability to CSF, and second, to what extent sufentanil is localized within the subarachnoid space to the level of the administration. These variables were studied in patients undergoing thoracotomy.
The study was approved by the Ethics Committee of the Medical Faculty at the University of Gothenburg and the Swedish Social Board of Welfare (Socialsty-relsen). Informed consent was obtained from each patient. Patients with liver or kidney disease, or neurologic disorder affecting the central nervous system or the spine were excluded. Thirty-seven patients undergoing elective thoracotomy were studied after allocation to either a lumbar (L) or thoracic (T) epidural group. Eight patients were excluded because of failures in CSF sampling. Of the remaining patients, 28 were operated for pulmonary tumors and one for spontaneous pneumothorax. There were no significant differences in sex, age, weight, duration of general anesthesia, or perioperative blood loss Table 1. Preanesthetic medication consisted of morphine hydrochloride 10 mg and scopolamine hydrobromide 0.4 mg intramuscularly. With the patient awake in the lateral position, a catheter (external diameter 0.85 mm and dead space 150 micro Liter) was inserted through an 18-gauge Touhy-Flower's needle into the subarachnoid space at the lumbar interspace L3-4 or L4-5 for CSF sampling. For sufentanil administration the same type of needle and catheter was inserted into the epidural space either in the lumbar interspace L2-3 or L3-4 or the thoracic interspace T5-6 or T6-7. The epidural space was identified with the "loss of resistance technique." A radial artery catheter was inserted for blood sampling.
The anesthesia was induced with IV thiopenthal 3-5 mg/kg followed by pancuronium 0.1 mg/kg IV. After orotracheal intubation anesthesia was continued with nitrous oxide in oxygen in proportions to maintain SaO2 at more than 95%. Volatile anesthetics were used if the proportion of nitrous oxide was less than 70%. Incremental doses of fentanyl 0.05-0.1 mg and pancuronium 0.5-2 mg were given IV when required throughout the operation. Shortly after the intubation, 75 micro gram sufentanil in 10 mL saline was administered according to allocation through the lumbar or thoracic epidural catheter.
Blood samples (10 mL) were collected at 3, 5, 10, 15, 30, 45, 60, 75, 90, 120, 180, 240, 300, 360, 480, and 600 min after sufentanil administration in the lumbar (L-plasma) and thoracic (T-plasma) groups. Plasma was separated by centrifugation. Through the subarachnoidal catheter, CSF samples (1 mL) were collected at 3, 5, 10, 15, 30, 45, 60, 75, 90, 120, 240, 360, 480, and 600 min after sufentanil administration in the lumbar (L-CSF) and thoracic (T-CSF) groups. Plasma and CSF samples were stored at -20 degrees C until analyzed. Sufentanil concentrations were determined by radioimmunoassay after extraction as described in detail elsewhere . The lower limit of quantification was 0.020 ng/mL providing that 1-mL volumes of plasma or 1-mL CSF samples could be extracted. Intra- and interassay coefficients of variation over a sufentanil concentration range of 0.056 to 9.75 ng/mL were 5.1%-8.5%, and 8.5%-10.5%, respectively.
The terminal elimination half-lives were calculated by least-squares regression. The area under the concentration curve (AUC) was calculated in CSF (AUC (CSF)) and plasma (AUCplasma) by the linear trapezoidal method with extrapolation to infinity. The extrapolated area (AUCresidual) was calculated as the ratio of the last calculated concentration (Clast) and the slope of the terminal phase (beta). The mean residual AUCCSF and AUCplasma was 7% and 21%, respectively, in the lumbar group (L-AUCCSF, L-AUCplasma) and 12% and 40%, respectively, in the thoracic group (T-AUCCSF, T-AUCplasma). The maximum concentration of sufentanil (Cmax) measured in CSF and plasma was determined in each patient; tmax was defined individually as the time to reach Cmax. The availability to CSF of epidural sufentanil was defined as the fraction of the lumbar epidural dose that was found in CSF and calculated as: Equation 1 where AUC (169.48 ng h/mL) after intrathecal administration of 15 micro gram sufentanil was taken from a previous study .
Data are presented as mean +/- SD. Statistical analysis was performed using the Mann-Whitney U-test for between-groups testing and Wilcoxon signed rank test for within-groups testing. P < 0.05 was considered statistically significant.
Sufentanil appeared more rapidly (P < 0.01) and in higher maximum concentrations (P < 0.01) in plasma after lumbar epidural administration compared to the thoracic approach Figure 1, Figure 2Table 2. AUCplasma, however, did not differ between the thoracic and lumbar groups.
In the lumbar group, the maximum concentration of sufentanil appeared six times slower in L-CSF than in L-plasma (P < 0.01) Table 2Figure 1, Figure 3. Albeit more variable, the sufentanil concentrations were higher in L-CSF than in L-plasma. The L-AUC and the maximum concentration in L-CSF were 19 and 44 times higher than corresponding values in L-plasma. The L-CSF/L-plasma concentration ratio declined to 8 and 5 at 6 and 10 h, respectively, illustrating a more rapid decrease in sufentanil concentrations in CSF than in plasma. The fraction that was recovered in CSF after lumbar epidural administration was calculated to be 2.7%.
In the thoracic group Cmax in CSF was 8.2 times lower (P < 0.001) and AUC in CSF five times lower than in the lumbar group (P < 0.01) Table 2Figure 3, Figure 4.
As with other opioids administered epidurally, sufentanil appeared in higher concentrations in CSF than in plasma, but the CSF-to-plasma concentration ratio was lower than that reported for morphine and pethidine. After epidural morphine administration, the CSF concentrations were 100-200 times higher than in plasma . After pethidine administration, the ratio of the maximum concentrations in CSF and plasma was 56  compared to 44 with sufentanil in the present study. This may be explained by a low availability of sufentanil to CSF from the epidural space and/or a rapid clearance from CSF for sufentanil. For obvious ethical reasons, in patients the amount of epidural opioids (e.g., sufentanil) that is absorbed in CSF has to be calculated by relating to concentration data from other patients given sufentanil by the intrathecal route. Moreover, only CSF concentration data of epidural and intrathecal opioids after lumbar administration may be used. Regarding this limitation, the availability to CSF of epidural sufentanil was approximately 3%, compared with 2% and 10% for morphine and pethidine, respectively [9,10]. It has been speculated that there is an extensive uptake of sufentanil into the epidural fat which would be less for the less lipophilic opioids . Obviously, there is no simple relationship between lipophilicity, extent of dural penetration, and CSF/plasma concentrations. After epidural administration the sufentanil concentrations in CSF by far exceed those after IV administration supporting the concept that there is also a spinal site of action after epidural administration .
On the other hand, clearance from CSF of morphine, pethidine, and sufentanil corresponds with the lipophilic properties of these opioids, such that sufentanil is cleared most rapidly and morphine least rapidly [7,13-17]. This finding implies an inverse relationship between clearance properties of an opioid from CSF and derived CSF/plasma concentration ratios of that specific opioid.
Sufentanil disappeared more rapidly from CSF than from plasma after intrathecal administration  and a similar characteristic was found in the present study after epidural administration. The terminal elimination half-lives averaged 4.1 and 2.8 h in plasma and CSF, respectively. Despite the fact that the ratio of the maximum concentrations in CSF and plasma was 44, the sufentanil concentrations were only five times higher in CSF than in plasma 10 h after the epidural administration. Consequently, after multiple administrations the tendency for accumulation would probably be greater in plasma than in CSF.
The difference in plasma disposition between the thoracic and lumbar groups may reflect variations in amount of fat  and blood circulation in the thoracic and lumbar epidural space . Compared to the lumbar region, the thoracic space is narrow. Consequently, when injected epidurally in the thoracic region a greater surface area of epidural fat would be reached to bind the lipophilic sufentanil. Moreover, in analogy to lumbar surgery during thoracic surgery, the absorption to plasma could be delayed due to thoracic vasoconstriction in the epidural space.
The sufentanil concentrations in the CSF in the lumbar region were considerably lower after epidural administration in the thoracic region than in the lumbar region. Thus, in CSF, after thoracic administration, the maximum concentration and the AUC were only 12% and 21%, respectively, of that found in CSF after lumbar administration. Morphine was studied previously under similar conditions, and after thoracic administration the corresponding values for Cmax and AUC for morphine were 53% and 51%, respectively . The movement of sufentanil in CSF occurs both by diffusion and bulk flow movement of CSF. CSF is produced in the ventricular system, and absorption occurs through arachnoid villi in both the cranial and spinal cavities inducing both a tailward and a headward flow of spinal fluid . The data on sufentanil and morphine show the extent of distribution within the subarachnoid space in a caudal direction. However, with regard to the bidirectional flow of spinal fluid, the data in the present study indicate also the amount of rostral distribution. Thus, the concentration gradient within CSF is more pronounced with the lipophilic sufentanil as compared to the hydrophilic morphine. Consequently, based on this pharmacokinetic knowledge, the tailoring of the epidural site of administration to that of surgical incision should be more important when administering a lipophilic drug, such as sufentanil, as compared to a hydrophilic drug, such as morphine. With the steeper concentration gradient within the CSF of sufentanil compared to morphine, it is conceivable that less influence on supraspinal opioid receptor sites related to ventilation would be obtained as well. Aside from morphine, this characteristic should be of greater consequence for other commonly used opioids, as they show smaller enhancement in potency when administered epidurally rather than IV . The accurate choice of segmental level for administration of epidural opioids has been shown not to be crucial for the efficacy of morphine, but aside from morphine this option is less obvious . The study by Sandler et al.  suggests that epidural infusions of fentanyl administered in the lumbar region for postoperative pain after thoracotomy resulted in similar analgesic, pharmacokinetic, and respiratory effects as after IV administration. However, with thoracic placement of the epidural catheter, fentanyl analgesia after thoracotomy is obtained with less dose requirements, plasma concentrations, and respiratory depression than after IV administration . In a recent study with sufentanil, a clinical advantage of thoracic over lumbar epidural administration after thoracotomy was demonstrated as well . These findings support the pharmacokinetic concept that, for lipophilic opioids, the site of epidural administration is of importance for optimum and safe analgesia.
In conclusion, we have shown that the concentrations of epidural sufentanil are higher in CSF than those in plasma, which supports the concept of a spinal site of action. However, in view of the high lipophilic properties of sufentanil, the availability to CSF from the epidural site of administration was less, and, consequently, the CSF concentrations were lower than anticipated. Sufentanil was found to be highly localized in CSF to the level of administration. This finding should influence our management of this technique in terms of tailoring the epidural site of administration to the level of nociceptive input. Moreover, sufentanil was eliminated more rapidly from CSF than from plasma. This observation implies a higher risk of accumulation of sufentanil in plasma than in CSF after multiple epidural administration.
1. Bromage P, Camporesi E, Durant P, et al. Rostral spread of epidural morphine. Anesthesiology 1982;56:431-6.
2. Gourlay G, Cherry D, Plummer J, et al. The influence of drug polarity on the absorption of opioid drugs into CSF and subsequent cephalad migration following lumbar epidural administration: application to morphine and pethidine. Pain 1987;31:297-305.
3. Gourlay GK, Murphy T, Plummer J, et al. Pharmacokinetics of fentanyl in lumbar and cervical CSF following lumbar epidural and intravenous administration. Pain 1989;38:253-9.
4. Van-Der-Auwera D, Verborgh C, Camu F. Analgesic and cardiorespiratory effects of sufentanil and morphine in humans. Anesth Analg 1987;66:999-1003.
5. Nordberg G. Pharmacokinetic aspects of spinal morphine analgesia. Acta Anaesthesiol Scand 1984;28:7-38.
6. Woestenborghs R, Timmerman P, Cornelissen M-L, et al. Assay methods for sufentanil in plasma: radioimmunoassay versus gas chromatography-mass spectrometry. Anesthesiology 1994;80(3):666-70.
7. Hansdottir V, Hedner T, Woestenborghs R, et al. The CSF and plasma pharmacokinetics of sufentanil after intrathecal administration. Anesthesiology 1991;74:264-9.
8. Nordberg G, Hedner T, Mellstrand T, et al. Pharmacokinetic aspects of epidural morphine analgesia. Anesthesiology 1983;58:545-51.
9. Nordberg G, Hansdottir V, Bondeson U, et al. CSF and plasma pharmacokinetics of pethidine and norpethidine in man after epidural and intrathecal administration of pethidine. Eur J Clin Pharmacol 1988;34:625-31.
10. Nordberg G, Hedner T, Mellstrand T, et al. Pharmacokinetics of epidural morphine in man. Eur J Clin Pharmacol 1984;26:233-7.
11. Boersma F, Noorduin H, Pieters W, et al. Sufentanil concentration in the spinal cord after long term epidural infusion [abstract]. Anesthesiology 1989;71:A706.
12. Taverne RH, Ionescu TI, Nuyten ST. Comparative absorption and distribution pharmacokinetics of intravenous and epidural sufentanil for major abdominal surgery. Clin Pharmacokinet 1992;23:231-7.
13. Nordberg G, Hedner T, Mellstrand T, et al. Pharmacokinetic aspects of intrathecal morphine analgesia. Anesthesiology 1984;60:448-54.
14. Sjostrom S, Hertvig P, Persson M, et al. Pharmacokinetics of intrathecal morphine and meperidine in humans. Anesthesiology 1987;67:889-95.
15. Ionescu T, Drost R, Roelofs J, et al. The pharmacokinetics of intradural morphine in major abdominal surgery. Clin Pharmacokinet 1988;14:178-86.
16. Kobot H, Hand C, Moore R, et al. Intrathecal morphine and heroin in humans: six-hour drug levels in spinal fluid and plasma. Anesth Analg 1986;65:718-22.
17. Ionescu T, Taverne R, Houweling P, et al. Pharmacokinetic study of extradural and intrathecal sufentanil anaesthesia for major surgery. Br J Anaesth 1991;66:458-64.
18. Nordberg G, Hansdottir V, Kvist L, et al. Pharmacokinetics of different epidural sites of morphine administration. Eur J Pharmacol 1987;33:499-504.
19. Somjen G. Spinal fluids and ions. In: Davidoff RA, ed. Handbook of the spinal cord. New York: Marcel Dekker, 1983;329-80.
20. Chrubasik J, Chrubasik S, Martin E. The ideal epidural opioid-fact or fantasy. Eur J Anaesthesiol 1993;10:79-100.
21. Fromme GA, Steidl LJ, Danielson DR. Comparison of lumbar and thoracic epidural morphine for relief of postthoracotomy pain. Anesth Analg 1985;64:454-5.
22. Sandler A, Stringer D, Panos L, et al. A randomized, double-blind comparison of lumbar epidural and intravenous fentanyl infusions for postthoracotomy pain relief. Anesthesiology 1992;77:626-34.
23. Salomaki T, Laitinen J, Nuutinen L. A randomized double-blind comparison of epidural versus intravenous fentanyl infusion for analgesia after thoracotomy. Anesthesiology 1991;75:790-5.
24. Hansdottir V, Bake B, Nordberg G. Pharmacodynamic aspects of epidural sufentanil and bupivacain infusion after thoracotomy. Acta Anaesth Scand 1995;39. In press.